Re: [PATCH] x86: Convert shadow-paging to Kconfig

[Tim Deegan <tim@xen.org>]

At 09:46 +0000 on 19 Jan (1453196796), Ian Campbell wrote:
> On Mon, 2016-01-18 at 18:40 +0000, Andrew Cooper wrote:
> > Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com>
> 
> Does this have any impact on migration of either PV or HVM guests? What
> about nested virt?

Without shadow paging you can't do live migration of PV guests.  That
should go in the kconfig description. 

AFAICT nested virt depends on HAP rather than shadow.

> Are things which are defined in xen/arch/*/Rules.mk in this way
> overrideable from the old top-level .config or does one need to dive deeper
> to modify them? If it's not configurable from top-level .config today then
> I think it either needs a "depends EXPERT" or for the changelog to make a
> convincing argument why this should be made user selectable.

I'm all in favour of making this selectable -- people who don't need
the feature can avoid 7k LOC that way. :)  If it's not going to be
'EXPERT' maybe we could have some console logging, either in the
none.c stubs or just at boot time, to make it more obvious why the
user's live migration/non-HAP guest doesn't work.

Cheers,

Tim.